Prefer the mimemagic gem for content type detection

Prioritize MimeMagic over the `file` binary for content type detection
when it finds a match. Fall back to `file` if MimeMagic can't match
anything.

`file` incorrectly detects Open Office XML files (e.g., xlsx, docx) as
zip since they're implemented as zipped archives of xml files. MimeMagic
detects them properly.

lib/paperclip.rb

@@ -57,6 +57,8 @@ require 'paperclip/attachment_registry'
 require 'paperclip/filename_cleaner'
 require 'paperclip/rails_environment'
 require 'mime/types'
+require 'mimemagic'
+require 'mimemagic/overlay'
 require 'logger'
 require 'cocaine'
 
@@ -137,7 +139,7 @@ module Paperclip
     #     user.avatar.url # => "/avatars/23/normal_me.png"
     # * +keep_old_files+: Keep the existing attachment files (original + resized) from
     #   being automatically deleted when an attachment is cleared or updated. Defaults to +false+.
-    # * +preserve_files+: Keep the existing attachment files in all cases, even if the parent 
+    # * +preserve_files+: Keep the existing attachment files in all cases, even if the parent
     #   record is destroyed. Defaults to +false+.
     # * +whiny+: Will raise an error if Paperclip cannot post_process an uploaded file due
     #   to a command line error. This will override the global setting for this attachment.

lib/paperclip/content_type_detector.rb

@@ -33,7 +33,7 @@ module Paperclip
       elsif calculated_type_matches.any?
         calculated_type_matches.first
       else
-        type_from_file_command || SENSIBLE_DEFAULT
+        type_from_file_contents || SENSIBLE_DEFAULT
       end.to_s
     end
 
@@ -54,11 +54,24 @@ module Paperclip
     end
 
     def calculated_type_matches
-      possible_types.select{|content_type| content_type == type_from_file_command }
+      possible_types.select { |content_type| content_type == type_from_file_contents }
+    end
+
+    def type_from_file_contents
+      type = begin
+               type_from_mime_magic || type_from_file_command
+             rescue Errno::ENOENT => e
+               Paperclip.log("Error while determining content type: #{e}")
+               SENSIBLE_DEFAULT
+             end
     end
 
     def type_from_file_command
       @type_from_file_command ||= FileCommandContentTypeDetector.new(@filename).detect
     end
+
+    def type_from_mime_magic
+      @type_from_mime_magic ||= MimeMagic.by_magic(File.open(@filename)).try(:type)
+    end
   end
 end

lib/paperclip/file_command_content_type_detector.rb

@@ -14,19 +14,17 @@ module Paperclip
 
     def type_from_file_command
       type = begin
-        # On BSDs, `file` doesn't give a result code of 1 if the file doesn't exist.
-        Paperclip.run("file", "-b --mime :file", :file => @filename)
-      rescue Cocaine::CommandLineError => e
-        Paperclip.log("Error while determining content type: #{e}")
-        SENSIBLE_DEFAULT
-      end
+               # On BSDs, `file` doesn't give a result code of 1 if the file doesn't exist.
+               Paperclip.run("file", "-b --mime :file", file: @filename)
+             rescue Cocaine::CommandLineError => e
+               Paperclip.log("Error while determining content type: #{e}")
+               SENSIBLE_DEFAULT
+             end
 
       if type.nil? || type.match(/\(.*?\)/)
         type = SENSIBLE_DEFAULT
       end
       type.split(/[:;\s]+/)[0]
     end
-
   end
 end
-

paperclip.gemspec

@@ -24,6 +24,7 @@ Gem::Specification.new do |s|
   s.add_dependency('activesupport', '>= 3.0.0')
   s.add_dependency('cocaine', '~> 0.5.5')
   s.add_dependency('mime-types')
+  s.add_dependency('mimemagic', '0.3.0')
 
   s.add_development_dependency('activerecord', '>= 3.0.0')
   s.add_development_dependency('shoulda')

spec/paperclip/content_type_detector_spec.rb

 require 'spec_helper'
 
 describe Paperclip::ContentTypeDetector do
+  it 'returns a meaningful content type for open xml spreadsheets' do
+    file = File.new(fixture_file("empty.xlsx"))
+    assert_equal "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+      Paperclip::ContentTypeDetector.new(file.path).detect
+  end
+
   it 'gives a sensible default when the name is empty' do
     assert_equal "application/octet-stream", Paperclip::ContentTypeDetector.new("").detect
   end
@@ -13,7 +19,7 @@ describe Paperclip::ContentTypeDetector do
 
   it 'returns content type of file if it is an acceptable type' do
     MIME::Types.stubs(:type_for).returns([MIME::Type.new('application/mp4'), MIME::Type.new('video/mp4'), MIME::Type.new('audio/mp4')])
-    Paperclip.stubs(:run).returns("video/mp4")
+    Paperclip::ContentTypeDetector.any_instance.stubs(:type_from_file_contents).returns("video/mp4")
     @filename = "my_file.mp4"
     assert_equal "video/mp4", Paperclip::ContentTypeDetector.new(@filename).detect
   end

spec/paperclip/file_command_content_type_detector_spec.rb

@@ -24,4 +24,3 @@ describe Paperclip::FileCommandContentTypeDetector do
       Paperclip::FileCommandContentTypeDetector.new("windows").detect
   end
 end
-

spec/paperclip/io_adapters/abstract_adapter_spec.rb

@@ -9,11 +9,12 @@ describe Paperclip::AbstractAdapter do
     end
   end
 
-  context "content type from file command" do
+  context "content type from file contents" do
     before do
       @adapter = TestAdapter.new
       @adapter.stubs(:path).returns("image.png")
       Paperclip.stubs(:run).returns("image/png\n")
+      Paperclip::ContentTypeDetector.any_instance.stubs(:type_from_mime_magic).returns("image/png")
     end
 
     it "returns the content type without newline" do

spec/paperclip/io_adapters/file_adapter_spec.rb

@@ -73,10 +73,13 @@ describe Paperclip::FileAdapter do
         end
       end
 
-      context "file with content type derived from file command on *nix" do
+      context "file with content type derived from file contents on *nix" do
         before do
           MIME::Types.stubs(:type_for).returns([])
           Paperclip.stubs(:run).returns("application/vnd.ms-office\n")
+          Paperclip::ContentTypeDetector.any_instance
+            .stubs(:type_from_mime_magic).returns("application/vnd.ms-office")
+
           @subject = Paperclip.io_adapters.for(@file)
         end